Internal Robustness: systematic search for systematic bias in SN Ia data

TL;DR

This paper introduces a Bayesian method to systematically detect and identify systematic biases in Type Ia supernova data, enhancing the reliability of cosmological measurements.

Contribution

The paper presents a novel Bayesian tool that extends goodness-of-fit tests to identify and locate systematic contamination in supernova datasets.

Findings

Union2.1 data shows no significant systematics

The method detects systematics when including lower-quality supernovae

Tool achieves high confidence in identifying contaminated data

Abstract

A great deal of effort is currently being devoted to understanding, estimating and removing systematic errors in cosmological data. In the particular case of type Ia supernovae, systematics are starting to dominate the error budget. Here we propose a Bayesian tool for carrying out a systematic search for systematic contamination. This serves as an extension to the standard goodness-of-fit tests and allows not only to cross-check raw or processed data for the presence of systematics but also to pin-point the data that are most likely contaminated. We successfully test our tool with mock catalogues and conclude that the Union2.1 data do not possess a significant amount of systematics. Finally, we show that if one includes in Union2.1 the supernovae that originally failed the quality cuts, our tool signals the presence of systematics at over 3.8-sigma confidence level.